A method and system for intelligent traffic management

Abstract

The present application relates to the field of traffic management, and in particular to a method and system for intelligent traffic management. The method includes: obtaining push content, and determining push audience information based on the push content; obtaining push audience behavior information, and determining dynamic behavior information based on the push audience information and the behavior information; obtaining the current content information of the push audience, and determining the push information based on the push audience information, the dynamic behavior information and the current content information. The present application determines push audience information based on push content, and provides an accurate push target audience for the push of push content; accurately captures and determines push audience dynamic behavior information based on push audience information and behavior information, and further optimizes push content in combination with the current content information of the push audience to make it more in line with the current interests and scenario requirements of the push audience. It not only improves the efficiency of traffic distribution, but also significantly enhances the participation and satisfaction of the push audience.

Description

Technical Field

[0001] The present application relates to the field of traffic management, and in particular to a method and system for intelligent traffic management. Background Art

[0002] With the improvement of people's living standards and the rapid development of science and technology, when people use network devices to browse information, the system background obtains and analyzes the behavioral interests of the push audience and accurately pushes the information and content needed by the push audience, so that the push audience has a better experience in the process of information browsing and reading.

[0003] However, since existing behavioral analysis and information push technologies push relevant content based on the behavioral interests of the push audience, this push method cannot meet the actual needs of the push audience for pushing relevant content, resulting in an objective deviation between the push of relevant content and the actual push needs of the push audience for pushing relevant content, affecting the push audience's own experience and the service efficiency of the system. Summary of the invention

[0004] The present application provides a traffic intelligent management method and system to solve the above problems.

[0005] In a first aspect, the present application provides a method for intelligent traffic management, the method comprising:

[0006] Obtain push content, and determine push recipient information based on the push content;

[0007] Acquire behavior information of a push recipient, and determine dynamic behavior information based on the push recipient information and the behavior information;

[0008] The current content information of the push audience is obtained, and the push information is determined according to the push audience information, the dynamic behavior information and the current content information.

[0009] Through this solution, the target audience is determined by comprehensive analysis of the push content, and the dynamic preferences of the push audience are further refined based on their personal behavior information, thus achieving accurate user portrait construction. Combined with the current browsing content information of the push audience, the push strategy is dynamically adjusted and optimized to ensure that the pushed content is not only highly relevant to the push audience's interests, but also closely follows their immediate needs and status. This significantly improves the pertinence and effectiveness of content push, enhances the push audience experience, and promotes the efficient dissemination and conversion of information.

[0010] Optionally, the behavior information includes real-time behavior information and historical behavior information, and determining the dynamic behavior information according to the behavior information and the push audience information includes:

[0011] Determining a push type of the pushed content according to the pushed content;

[0012] Determining behavior difference information according to the real-time behavior information and the historical behavior information;

[0013] The dynamic behavior information of the push recipient is determined according to the behavior difference information, the push type and the push recipient information.

[0014] Through this solution, the push type is clarified according to the nature of the pushed content, providing a basis for subsequent personalized push. The real-time behavior information and historical behavior information of the push audience are comprehensively collected. By comparing and analyzing these two types of information, the behavioral difference information of the audience's behavior over time is revealed, which helps to capture the current interests and push needs of the push audience. Combining the behavioral difference information, push type and push audience information, a comprehensive evaluation is conducted to obtain the dynamic behavior information of the push audience, ensuring that the push strategy can closely fit the current preferences of the push audience for the pushed content, thereby greatly improving the pertinence and effectiveness of the push information, optimizing the push audience experience, and improving the accuracy of content push.

[0015] Optionally, the real-time behavior information includes browsing time, pop-up information, click information and viewing information of the push audience for the current content, and the historical behavior information includes historical attention information. The behavior difference information is determined based on the real-time behavior information and the historical behavior information, including:

[0016] Determining the bounce rate of the push audience for the current content according to the bounce information;

[0017] Determining the click rate of the push audience for the current content according to the click information;

[0018] Determining the viewing completeness of the current content by the push audience according to the viewing information;

[0019] Obtaining historical content and current content, and determining whether the current content and the historical content are of the same type;

[0020] If the current content and the historical content are of the same type, the current interest level of the push audience for the current content is determined based on the browsing time, the bounce rate, the click rate, and the viewing completeness, and is calculated according to the following formula:

[0021] ;

[0022] in, represents the current level of interest, Indicates the preset time weight, Indicates the browsing time. Indicates the preset bounce rate weight. represents the bounce rate, Indicates the preset click rate weight. represents the click-through rate, Indicates the preset completeness weight, Indicates the viewing completeness. Indicates preset adjustment constant;

[0023] Determining the historical interest level of the push recipient in the historically-focused content according to the historically-focused information;

[0024] Obtaining an interest level error range, and determining whether a difference between the historical interest level and the current interest level is within the interest level error range;

[0025] If the difference between the historical interest level and the current interest level is not within the interest level error range, determining the historical browsing time, historical bounce rate, historical click rate and historical viewing completeness according to the historical attention information;

[0026] Determine a browsing time difference, a bounce rate difference, a click rate difference and a viewing completeness difference according to the historical browsing time, the historical bounce rate, the historical click rate, the historical viewing completeness, the browsing time, the bounce rate, the click rate and the viewing completeness;

[0027] The browsing time difference, the bounce rate difference, the click rate difference and the viewing completeness difference are determined as the behavior difference information.

[0028] Through this solution, the depth and breadth of the push audience's interaction with the current content can be quantified based on key indicators such as bounce rate, click-through rate and viewing completeness. By comparing the current content with the historical content type, if they are consistent, a comprehensive formula is used to combine factors such as browsing time, bounce rate, click-through rate and viewing completeness to accurately calculate the current interest of the push audience in the current content. At the same time, refer to historical attention information to determine whether the push audience's interests have changed significantly. If it exceeds the preset error range, further analyze the specific differences between historical and current browsing behaviors, including the difference in browsing time, bounce rate, click-through rate and viewing completeness. This is used as behavioral difference information to provide a scientific basis for the adjustment and optimization of precise push strategies, ensuring that the push content is more in line with the actual interests and needs of the push audience.

[0029] Optionally, the push audience information includes an active period of the push audience, and determining the dynamic behavior information of the push audience according to the behavior difference information, the push type and the push audience information includes:

[0030] According to the browsing time difference, the bounce rate difference, the click rate difference and the viewing completeness difference, a comprehensive difference index is determined and calculated according to the following formula:

[0031] ;

[0032] in, represents the comprehensive difference index, Indicates the preset browsing time coefficient. represents the browsing time difference, Indicates the preset bounce rate coefficient. represents the bounce rate difference, Indicates the preset click rate coefficient, represents the click rate difference, Indicates the preset completeness coefficient, represents the viewing completeness difference;

[0033] Determining the current attention time and historical attention time of the push audience to the current content according to the comprehensive difference index;

[0034] The current attention time, the historical attention time and the active time period are determined as the dynamic behavior information.

[0035] Through this solution, by combining the browsing time difference, bounce rate difference, click rate difference and viewing completeness difference, and assigning each difference with a preset coefficient, a comprehensive difference index is calculated, which effectively measures the comprehensive difference between the current and historical behaviors of the push audience. Based on the analysis of the comprehensive difference index, the current attention time and historical attention time of the push audience on the current content are accurately determined. The comparison of these two time points reveals the changing trend of the push audience's interest in browsing content. Combining the current attention time, historical attention time and the active period of the push audience forms complete dynamic behavior information, which provides strong data support for in-depth understanding of the push audience's behavior patterns and optimization of push strategies, and helps to improve the accuracy of content push and the push audience's participation.

[0036] Optionally, the current content information includes a search type of the current content, and the determining the push information according to the push audience information, the dynamic behavior information and the current content information includes:

[0037] Determine the push frequency according to the push type, the push recipient, the current interest level and the search type;

[0038] Determine a push time period according to the active time period and the current attention time;

[0039] Determine push balance information according to the historical attention time, the current attention time, the historical content and the current content;

[0040] The push frequency, the push time period, and the push balance information are determined as the push information.

[0041] Through this solution, the push frequency is accurately determined through intelligent algorithms to ensure that the pushed content neither disturbs the push audience too frequently nor misses the best display opportunity due to sparseness. At the same time, according to the active time period and current attention time of the push audience, the push time period is accurately set so that the push content can be presented at the time point when the push audience is most likely to pay attention, thereby improving the effectiveness of the push content. In addition, combined with a comprehensive analysis of historical attention time, current attention time, historical content and current content, push balance information is formulated to maintain the freshness and coherence of the push content and avoid repetition or disconnection among the push audience. The push frequency, push time period and push balance information are integrated into push information to achieve comprehensive optimization of push information, ensure that the push of push content is both efficient and personalized, and effectively improve the satisfaction and participation of the push audience.

[0042] Optionally, determining the push frequency according to the push type, the push recipient, the interest level, and the search type includes:

[0043] Determining, according to the current content and the historical content, a current content type corresponding to the current content and a historical content type corresponding to the historical content;

[0044] Determining a content similarity degree according to the current content type and the historical content type;

[0045] According to the similarity of the contents, determining whether the push type is consistent with the search type;

[0046] If the push type and the search type are consistent, the interest level level is obtained, the current interest level is matched with the interest level level, and a push frequency corresponding to the current interest level is obtained.

[0047] Through this solution, by comparing and analyzing the current content and historical content, the types of each are clarified, and the similarity of the content between the two is evaluated accordingly, providing a basis for the formulation of subsequent push strategies. Based on the similarity of the content, it is determined whether the push type is consistent with the search type to ensure that the push content is highly consistent with the push audience's intention. If consistent, the interest level system is further introduced to match the current interest level with the interest level level, and the push frequency is determined based on this to ensure that the push intensity matches the push audience's interest in the push content. While ensuring the push audience experience, it effectively improves the visibility and conversion rate of push information, and realizes the seamless connection between the precise push of push content and the push audience's demand for push content.

[0048] Optionally, determining the push time period according to the active time period and the current attention time includes:

[0049] According to the active time period and the current attention time, determining a number of time distribution nodes at which the push audience pays attention to the current content within the active time period;

[0050] Acquire a number of time distribution coordinates of the number of time distribution nodes in one day and the total browsing time of the push audience in one day, and determine the attention intensity of the push audience to the current content according to the total browsing time and the current attention time;

[0051] According to the time distribution coordinates, the attention intensity, the total duration of a day and the active period, the node time ratio corresponding to each time distribution node is determined and calculated according to the following formula:

[0052] ;

[0053] in, Indicates The active duration of each time node, represents the intensity of attention, Indicates The location of the time node, represents the total length of the day in question;

[0054] The focus position of the push audience when paying attention to the current content is obtained, and the push time period is determined according to the node time ratio and the focus position.

[0055] Through this solution, by identifying several time distribution nodes, the audience's attention timeline can be accurately depicted. Furthermore, by combining the time distribution coordinates of these nodes in a day and the total browsing time of the push audience, the push audience's attention intensity to the current content is calculated, which effectively measures the push audience's interest and involvement in the current content. Using information such as attention intensity, time node location, total duration of a day, and active time period, the active time ratio of each time node is calculated through a specific formula, reflecting the activity level of the push audience at different time points. Combined with the location information of the push audience when paying attention and the node time ratio, the push time period is accurately determined to ensure that the push information can be presented at the time point when the push audience is most active and concerned, thereby significantly improving the push effect and push audience participation.

[0056] Optionally, the determining push balance information according to the historical attention time, the current attention time, the historical content and the current content includes:

[0057] Determining a first consumed time corresponding to the historical content according to the historical attention time and the historical content;

[0058] Determine, according to the current attention time and the current content, the current consumption time corresponding to the current content;

[0059] According to the consumed time, the first consumed time and the current consumed time, determine the total current push time of the current content within one day and the total historical push time of the historical content within one day, and determine the current push time and the historical push time as the push balance information.

[0060] Through this solution, based on the historical attention time and historical content, the first consumption time consumed by the historical content is calculated, reflecting the degree of attention of the push audience to similar content in the past. Combined with the current attention time and the current content, the current consumption time corresponding to the current content is determined, which effectively measures the push audience's immediate interest and attention to the current content. By comparing the historical consumption time, the first consumption time and the current consumption time, the total push time of the current content in a day and the total push time of the historical content in a day are comprehensively evaluated to ensure the balance of the push content and avoid the push audience from feeling fatigued or ignoring other important information due to excessive reception of similar information. Using the current push time and the historical push time as push balance information provides an important basis for formulating scientific and reasonable push strategies, which helps to improve the satisfaction of the push audience and the dissemination effect of the push content.

[0061] Optionally, after determining the dynamic behavior information of the push recipient according to the behavior difference information and the push recipient information, the method further includes:

[0062] Obtaining the number of views and first viewing completeness of a number of contents corresponding to the current content type by the push audience after the push;

[0063] Determining the same characteristic information of the plurality of contents according to the number of views corresponding to each content and the first viewing completeness;

[0064] Obtaining a number of dwell times and a number of click rates of a number of contents corresponding to the same characteristic information;

[0065] Determining, based on the plurality of dwell times and the plurality of click rates, the attractiveness of the plurality of contents corresponding to the same characteristic information to the push audience;

[0066] According to the attraction degree, it is determined whether the push recipient has a long-term interest in the plurality of contents corresponding to the same characteristic information.

[0067] Through this solution, by comprehensively tracking the browsing behavior of push audiences on content related to the current content type after receiving the push content, including the number of views and the completeness of the first viewing, the push audience's acceptance of these contents is deeply analyzed. Furthermore, based on the same feature information extracted from the browsing behavior, the dwell time and click-through rate of the corresponding content are recorded in detail. These two indicators together reflect the attractiveness of the content to the push audience. Through the comprehensive analysis of these data, it is accurately assessed which features of content have continued appeal to the push audience, so as to determine whether the push audience has a long-term interest in these feature contents. It not only helps to optimize the selection and strategy of push content, but also provides valuable feedback to content creators, promotes continuous innovation and improvement of push content, and meets the long-term interests and needs of push audiences.

[0068] In a second aspect, the present application provides a traffic intelligent management system, the system comprising:

[0069] An audience information determination module is used to obtain push content and determine push audience information based on the push content;

[0070] A behavior information determination module, used to obtain the behavior information of the push recipient, and determine the dynamic behavior information according to the push recipient information and the behavior information;

[0071] The push information determination module is used to obtain the current content information of the push audience, and determine the push information according to the push audience information, the dynamic behavior information and the current content information.

[0072] Optionally, the behavior information determination module is used to:

[0073] Determining a push type of the pushed content according to the pushed content;

[0074] Determining behavior difference information according to the real-time behavior information and the historical behavior information;

[0075] The dynamic behavior information of the push recipient is determined according to the behavior difference information, the push type and the push recipient information.

[0076] Optionally, the behavior information determination module is used to:

[0077] Determining the bounce rate of the push audience for the current content according to the bounce information;

[0078] Determining the click rate of the push audience for the current content according to the click information;

[0079] Determining the viewing completeness of the current content by the push audience according to the viewing information;

[0080] Obtaining historical content and current content, and determining whether the current content and the historical content are of the same type;

[0081] If the current content and the historical content are of the same type, the current interest level of the push audience for the current content is determined based on the browsing time, the bounce rate, the click rate, and the viewing completeness, and is calculated according to the following formula:

[0082] ;

[0083] in, represents the current level of interest, Indicates the preset time weight, Indicates the browsing time. Indicates the preset bounce rate weight. represents the bounce rate, Indicates the preset click rate weight. represents the click-through rate, Indicates the preset completeness weight, Indicates the viewing completeness. Indicates preset adjustment constant;

[0084] Determining the historical interest level of the push recipient in the historically-focused content according to the historically-focused information;

[0085] Obtaining an interest level error range, and determining whether a difference between the historical interest level and the current interest level is within the interest level error range;

[0086] If the difference between the historical interest level and the current interest level is not within the interest level error range, determining the historical browsing time, historical bounce rate, historical click rate and historical viewing completeness according to the historical attention information;

[0087] Determine a browsing time difference, a bounce rate difference, a click rate difference and a viewing completeness difference according to the historical browsing time, the historical bounce rate, the historical click rate, the historical viewing completeness, the browsing time, the bounce rate, the click rate and the viewing completeness;

[0088] The browsing time difference, the bounce rate difference, the click rate difference and the viewing completeness difference are determined as the behavior difference information.

[0089] Optionally, the behavior information determination module is used to:

[0090] According to the browsing time difference, the bounce rate difference, the click rate difference and the viewing completeness difference, a comprehensive difference index is determined and calculated according to the following formula:

[0091] ;

[0092] in, represents the comprehensive difference index, Indicates the preset browsing time coefficient. represents the browsing time difference, Indicates the preset bounce rate coefficient. represents the bounce rate difference, Indicates the preset click rate coefficient, represents the click rate difference, Indicates the preset completeness coefficient, represents the viewing completeness difference;

[0093] Determining the current attention time and historical attention time of the push audience to the current content according to the comprehensive difference index;

[0094] The current attention time, the historical attention time and the active time period are determined as the dynamic behavior information.

[0095] Optionally, the push information determination module is used to:

[0096] Determine the push frequency according to the push type, the push recipient, the current interest level and the search type;

[0097] Determine a push time period according to the active time period and the current attention time;

[0098] Determine push balance information according to the historical attention time, the current attention time, the historical content and the current content;

[0099] The push frequency, the push time period, and the push balance information are determined as the push information.

[0100] Optionally, the push information determination module is used to:

[0101] Determining, according to the current content and the historical content, a current content type corresponding to the current content and a historical content type corresponding to the historical content;

[0102] Determining a content similarity degree according to the current content type and the historical content type;

[0103] According to the similarity of the contents, determining whether the push type is consistent with the search type;

[0104] If the push type and the search type are consistent, the interest level level is obtained, the current interest level is matched with the interest level level, and a push frequency corresponding to the current interest level is obtained.

[0105] Optionally, the push information determination module is used to:

[0106] According to the active time period and the current attention time, determining a number of time distribution nodes at which the push audience pays attention to the current content within the active time period;

[0107] Acquire a number of time distribution coordinates of the number of time distribution nodes in one day and the total browsing time of the push audience in one day, and determine the attention intensity of the push audience to the current content according to the total browsing time and the current attention time;

[0108] According to the time distribution coordinates, the attention intensity, the total duration of a day and the active period, the node time ratio corresponding to each time distribution node is determined and calculated according to the following formula:

[0109] ;

[0110] in, Indicates The active duration of each time node, represents the intensity of attention, Indicates The location of the time node, represents the total length of the day in question;

[0111] The focus position of the push audience when paying attention to the current content is obtained, and the push time period is determined according to the node time ratio and the focus position.

[0112] Optionally, the push information determination module is used to:

[0113] Determining a first consumed time corresponding to the historical content according to the historical attention time and the historical content;

[0114] Determine, according to the current attention time and the current content, the current consumption time corresponding to the current content;

[0115] According to the consumed time, the first consumed time and the current consumed time, determine the total current push time of the current content within one day and the total historical push time of the historical content within one day, and determine the current push time and the historical push time as the push balance information.

[0116] Optionally, the intelligent traffic management system further includes a long-term interest determination module, which is used to:

[0117] Obtaining the number of views and first viewing completeness of a number of contents corresponding to the current content type by the push audience after the push;

[0118] Determining the same characteristic information of the plurality of contents according to the number of views corresponding to each content and the first viewing completeness;

[0119] Obtaining a number of dwell times and a number of click rates of a number of contents corresponding to the same characteristic information;

[0120] Determining, based on the plurality of dwell times and the plurality of click rates, the attractiveness of the plurality of contents corresponding to the same characteristic information to the push audience;

[0121] According to the attraction degree, it is determined whether the push recipient has a long-term interest in the plurality of contents corresponding to the same characteristic information. BRIEF DESCRIPTION OF THE DRAWINGS

[0122] In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, a brief introduction will be given below to the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative labor.

[0123] Figure 1 A schematic diagram of an application scenario provided for an embodiment of the present application;

[0124] Figure 2 A flow chart of a method for intelligent traffic management provided in one embodiment of the present application;

[0125] Figure 3 A schematic diagram of the structure of an intelligent traffic management system provided in one embodiment of the present application. DETAILED DESCRIPTION

[0126] In order to make the purpose, technical scheme and advantages of the embodiments of the present application clearer, the technical scheme in the embodiments of the present application will be clearly and completely described below in conjunction with the drawings in the embodiments of the present application. Obviously, the described embodiments are part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of this application.

[0127] In addition, the term "and / or" in this article is only a description of the association relationship of associated objects, indicating that there can be three relationships. For example, A and / or B can represent: A exists alone, A and B exist at the same time, and B exists alone. In addition, the character " / " in this article, unless otherwise specified, generally means that the associated objects before and after are in an "or" relationship.

[0128] The embodiments of the present application are further described in detail below in conjunction with the drawings in the specification.

[0129] Since existing behavioral analysis and information push technologies push relevant content based on the behavioral interests of the push audience, this push method cannot meet the actual needs of the push audience for the push of relevant content, resulting in an objective deviation between the push of relevant content and the actual push needs of the push audience for the push of relevant content, affecting the push audience's own experience and the service efficiency of the system.

[0130] Based on this, the present application provides a method and system for intelligent traffic management, which determines the target audience by comprehensively analyzing the pushed content, and further refines the dynamic preferences of the pushed audience based on their personal behavior information, thereby achieving accurate user portrait construction. Combined with the current browsing content information of the pushed audience, the push strategy is dynamically adjusted and optimized to ensure that the pushed content is not only highly relevant to the interests of the pushed audience, but also closely follows their immediate needs and status. It significantly improves the pertinence and effectiveness of content push, which not only enhances the push audience experience, but also promotes the efficient dissemination and conversion of information.

[0131] Figure 1 A schematic diagram of an application scenario provided for this application. When managing the traffic of pushed content, the method provided by this application is applied to obtain and analyze the behavior information, current content information and pushed content of the push audience, so as to accurately determine the push information that fits the actual experience needs of the push audience.

[0132] Specifically, the method provided in the present application is applied to any server, and the server interacts with the push management system, which can be a system for managing and controlling push content and push services. The server obtains and analyzes the push content, behavior information, and current content information in the push management system, thereby determining the push information and sending the push information to the push management system. By obtaining the push content, behavior information, and current content information, an accurate and comprehensive basis for determining the subsequent push information is provided, so that the push management system's push is more in line with the actual needs of the push audience, and the push audience's experience is improved.

[0133] For specific implementation methods, please refer to the following embodiments.

[0134] Figure 2 This is a flow chart of a method for intelligent traffic management provided by an embodiment of the present application. The method of this embodiment can be applied to the server in the above scenario. Figure 2 As shown, the method includes:

[0135] S201: Acquire push content, and determine push audience information based on the push content.

[0136] The pushed content may be content determined when pushing to a push recipient.

[0137] The push audience information may be information of a corresponding audience determined according to the push content.

[0138] Specifically, before pushing the push content, in order to ensure the accuracy of pushing the push content, it is necessary to analyze the push content to be pushed before pushing the push content to determine the target audience of the push content.

[0139] Obtain the target audience features corresponding to the keywords of the push content to be pushed and all possible push content from the push management system, integrate the target audience features corresponding to the keywords of all possible push content, and obtain a preset keyword feature set. Use natural language processing technology to analyze the push content to obtain the push keywords in the push content. Match the push keywords with the preset keyword feature set to obtain the target audience features corresponding to the push keywords, and determine the target audience features obtained in the above steps as the push audience information.

[0140] S202: Obtain behavior information of the push recipient, and determine dynamic behavior information based on the push recipient information and the behavior information.

[0141] The behavior information may be information corresponding to the behavior generated when the push audience browses historical content and current content.

[0142] Dynamic behavior information may be information generated when a push audience performs a specific operation on the content before the push content is pushed to the push audience.

[0143] Specifically, before the push content is pushed to the push audience, the push audience's reaction level and operation to browsing the content are in a real-time changing state. At this time, it is necessary to determine the push audience's dynamic behavior information on browsing the content based on the push audience's operation behavior on the content and the push audience's own information, so as to provide an accurate reference basis for the determination of subsequent push information.

[0144] The age, gender and education level of the push audience are extracted from the push audience information. The push content is processed using natural language processing technology to obtain the word count and push platform of the push content. Based on the push platform and the word count of the push content obtained in the above steps, the content browsed by the push audience before reading the push content is obtained from the push management system. The age, gender, education level and the content browsed by the push audience before reading the push content obtained in the above steps are analyzed using the stream processing framework technology to obtain the browsing time, click-through rate and bounce rate of the push audience for the content browsed by the push audience before reading the push content. The browsing time, click-through rate and bounce rate of the push audience for the content browsed by the push audience before reading the push content are determined as dynamic behavior information.

[0145] S203: Acquire current content information of the push recipient, and determine push information according to the push recipient information, the dynamic behavior information and the current content information.

[0146] The current content information may be information corresponding to the content that the push audience is paying attention to when browsing and paying attention to the content.

[0147] The push information may be information related to the push determined when the content is pushed to the push recipient. The push information may include the push frequency, push period, and push duration.

[0148] Specifically, when the push audience browses the content, the push audience's attention to the content will also change accordingly. Therefore, when determining the push information, it is necessary to consider the dynamic behavior of the push audience, the push audience's own relevant information, and the changes in the push audience's specific information on the current content of concern, so as to determine the push information.

[0149] Obtain the content information that the push audience is paying attention to and the browsing time, click-through rate and bounce rate of the content information that the push audience is paying attention to from the push management system, and determine the content information that the push audience is paying attention to as the current content information. Extract the type of the current content from the current content information. Based on the type of the current content obtained in the above steps, use the time series prediction technology to analyze the browsing time of the current content by the push audience obtained in the above steps to obtain the push time period. Use the reinforcement learning technology to analyze the click-through rate, bounce rate, age, gender and education level of the push audience obtained in the above steps to obtain the push frequency and push duration. Determine the push frequency, push duration and push time period as the push information.

[0150] Through this solution, the target audience is determined by comprehensive analysis of the push content, and the dynamic preferences of the push audience are further refined based on their personal behavior information, thus achieving accurate user portrait construction. Combined with the current browsing content information of the push audience, the push strategy is dynamically adjusted and optimized to ensure that the pushed content is not only highly relevant to the push audience's interests, but also closely follows their immediate needs and status. This significantly improves the pertinence and effectiveness of content push, enhances the push audience experience, and promotes the efficient dissemination and conversion of information.

[0151] In some embodiments, the push type of the pushed content is determined based on the pushed content; the behavior difference information is determined based on the real-time behavior information and the historical behavior information; the dynamic behavior information of the push audience is determined based on the behavior difference information, the push type and the push audience information.

[0152] The push type may be a type corresponding to the pushed content.

[0153] Real-time behavior information can be information generated when the push audience performs operations on the content while paying attention to the content.

[0154] The historical behavior information may be information generated when the push audience performs operations on the content when paying attention to the historical content.

[0155] The behavior difference information may be information about the difference between the operation behaviors performed by the push audience on the historical content and the current content.

[0156] Specifically, when the push audience pays attention to content, the push audience's attention behavior to different content at different times is different. Therefore, there will be certain differences in the push audience's attention behavior corresponding to the content currently paid attention to and the content paid attention to in the past. At this time, it is necessary to obtain and analyze the push audience's behavior information on the current content and the behavior information on the historical content, so as to determine the difference between the push audience's operating behaviors on the historical content and the current content, and provide an accurate reference basis for the determination of subsequent dynamic behavior information.

[0157] Obtain all possible keywords and their corresponding types of push content from the push management system, integrate all possible keywords and their corresponding types of push content, and obtain a preset keyword type set. Use natural language processing technology to analyze the push content to obtain the keywords of the push content. Match the keywords of the push content with the preset keyword type set to obtain the push type corresponding to the push content. Extract the real-time behavior information and historical behavior information of the push audience from the behavior information. Extract the real-time browsing time, real-time click-through rate and real-time bounce rate from the real-time behavior information. Extract the historical browsing time, historical click-through rate and historical bounce rate from the historical behavior information. Subtract the real-time browsing time, real-time click-through rate and real-time bounce rate from the historical browsing time, historical click-through rate and historical bounce rate respectively to obtain the browsing time difference, click-through rate difference and bounce rate difference. Determine the browsing time difference, click-through rate difference and bounce rate difference as behavior difference information. Based on the push type obtained in the above steps, the total time of 24 hours a day is divided into eight browsing periods, and the corresponding time of each browsing period is 3 hours. According to the browsing time difference, click rate difference and bounce rate difference obtained in the above steps, the historical browsing time, historical click rate, historical bounce rate, real-time click rate, real-time bounce rate and real-time browsing time corresponding to each browsing period are determined. The historical browsing time, historical click rate, historical bounce rate, real-time click rate, real-time bounce rate and real-time browsing time corresponding to each browsing period are determined as dynamic behavior information.

[0158] Through this solution, the push type is clarified according to the nature of the pushed content, providing a basis for subsequent personalized push. The real-time behavior information and historical behavior information of the push audience are comprehensively collected. By comparing and analyzing these two types of information, the behavioral difference information of the audience's behavior over time is revealed, which helps to capture the current interests and push needs of the push audience. Combining the behavioral difference information, push type and push audience information, a comprehensive evaluation is conducted to obtain the dynamic behavior information of the push audience, ensuring that the push strategy can closely fit the current preferences of the push audience for the pushed content, thereby greatly improving the pertinence and effectiveness of the push information, optimizing the push audience experience, and improving the accuracy of content push.

[0159] In some embodiments, the bounce rate of the push audience for the current content is determined based on the bounce information; the click rate of the push audience for the current content is determined based on the click information; the viewing completeness of the push audience for the current content is determined based on the viewing information; the historical content and the current content are obtained to determine whether the current content and the historical content are of the same type; if the current content and the historical content are of the same type, the current interest level of the push audience for the current content is determined based on the browsing time, the bounce rate, the click rate and the viewing completeness according to formula (1):

[0160] (1)

[0161] in, Indicates the current level of interest, Indicates the preset time weight, Indicates browsing time. Indicates the preset bounce rate weight. Represents the bounce rate. Indicates the preset click rate weight. Indicates the click rate, Indicates the preset completeness weight, Indicates viewing completeness. Indicates preset adjustment constant;

[0162] Based on the historical attention information, determine the historical interest level of the push audience in the historical attention content; obtain the interest level error range, and determine whether the difference between the historical interest level and the current interest level is within the interest level error range; if the difference between the historical interest level and the current interest level is not within the interest level error range, then determine the historical browsing time, historical bounce rate, historical click-through rate and historical viewing completeness based on the historical attention information; determine the browsing time difference, bounce rate difference, click-through rate difference and viewing completeness difference based on the historical browsing time, historical bounce rate, historical click-through rate, historical viewing completeness, browsing time, bounce rate, click-through rate and viewing completeness; determine the browsing time difference, bounce rate difference, click-through rate difference and viewing completeness difference as behavior difference information.

[0163] The pop-up information may be information corresponding to when the push audience performs a pop-up operation on the page of the current content when following the current content being followed.

[0164] The click information may be information corresponding to when the push audience clicks on the page of the current content when paying attention to the current content being paid attention to.

[0165] The viewing information may be information corresponding to when the push audience views the current content.

[0166] The bounce rate may be a ratio between the number of visits in which the push audience leaves after browsing the current content and the total number of visits.

[0167] The click-through rate may be a ratio of the number of clicks on the current content by the push audience to the total number of impressions.

[0168] The viewing completeness may be a ratio of the viewing time of the current content to the total viewing time of the push audience or a ratio of the viewed portion of the current content to the total content portion.

[0169] The current interest level may indicate the degree of attention, importance or preference of the push recipient to the current content.

[0170] The historical interest level may indicate the degree of attention, importance or preference of the push recipients for the current content.

[0171] The interest level error range may be a range that measures whether the content type corresponding to the current interest level and the content type corresponding to the historical interest level are the same type.

[0172] The preset time weight may be a numerical value that measures the degree to which the browsing time affects the calculation result when calculating the current interest level of the push audience.

[0173] The preset bounce rate weight may be a numerical value that measures the degree to which the bounce rate affects the calculation result when calculating the current interest level of the push audience.

[0174] The preset completeness weight may be a numerical value that measures the degree to which the viewing completeness affects the calculation result when calculating the current interest level of the push audience.

[0175] The preset adjustment constant may be a value used to correct the viewing completeness value when calculating the current interest level of the push audience so that the determination of the current interest level conforms to the actual situation.

[0176] Specifically, when the push audience pays attention to content, there will be certain differences in some specific operational behaviors of the push audience towards historical content and current content. In order to distinguish the differences in specific operational behaviors between historical content and current content, it is necessary to determine the degree of behavioral interest of the push audience in historical content and current content based on the push audience's attention operational behaviors towards historical content and current content, so as to provide an accurate reference basis for determining subsequent behavioral difference information.

[0177] Extract the number of visits and the total number of visits that the push audience has left after browsing the current content from the bounce information, and divide the number of visits that the push audience has left after browsing the current content by the total number of visits to obtain the bounce rate. Extract the number of clicks and the total number of impressions of the push audience on the current content from the click information, and divide the number of clicks and the total number of impressions of the push audience on the current content to obtain the click rate. Extract the viewing time of the push audience on the current content and the total time of the current content or the browsed part of the current content and the total content part from the viewing information, and divide the viewing time of the push audience on the current content by the total time of the current content or divide the browsed part of the current content by the total content part to obtain the viewing completeness. Obtain historical content and current content from the push management system, use natural language processing technology to analyze the historical content and current content, and obtain keywords of the historical content and keywords of the current content. Match the keywords of the historical content and the keywords of the current content with the preset keyword type set respectively to obtain the type corresponding to the historical content and the type corresponding to the current content. Compare the type corresponding to the historical content with the type corresponding to the current content to determine whether the type corresponding to the historical content is consistent with the type corresponding to the current content. If they are consistent, obtain the historical viewing completeness and historical interest level from the push management system, and use the linear regression analysis method to analyze the historical viewing completeness, historical click-through rate, historical bounce rate, historical browsing time and historical interest level to obtain the preset time weight, preset bounce rate weight, preset click rate weight, preset completeness weight and preset adjustment constant. Use the mathematical analysis method to analyze the preset time weight, preset bounce rate weight, preset click rate weight, preset completeness weight, preset adjustment constant, browsing time, bounce rate, click rate and viewing completeness obtained in the above steps, and calculate according to formula (1) to obtain the current interest level of the push audience in the current content.

[0178] In formula (1), the impact of browsing time does not increase infinitely. For example, the difference between a push audience staying for 10 seconds and staying for 100 seconds will not be amplified like a linear function, but will increase the interest score relatively smoothly. The logarithmic function ensures that the "marginal impact" of long-term stay is not too exaggerated. The longer the push audience stays on the page, it usually means that they are more interested in the content. This part reflects the positive contribution of browsing time to the interest score, but as time goes by, the increase in the interest score gradually decreases, avoiding over-emphasizing the situation where the push audience stays on the content page for a long time.

[0179] Represents the probability of no bounce, indicating that the more active the push audience is on the page, the lower the bounce rate, and the higher the push audience's interest in the page. Use the square function The purpose is to emphasize the impact of a low bounce rate. That is, when the bounce rate is low, the interest score will increase, and when the bounce rate is high, the interest score will decrease significantly. This nonlinear relationship avoids overemphasizing the slight effects that a simple linear calculation might produce.

[0180] The impact of click-through rate is gradually decreasing. Even if the click-through rate is very high, the square root function will reduce the excessive weighting of extremely high click-through rates on the interest score. This can avoid the bias caused by "too many clicks".

[0181] The push audience clicked more content on the page, which means they have a higher interest in the page content, so this part increases the interest score. Square root function This ensures that the impact of click-through rate on the score decreases smoothly, avoiding excessive amplification when there are too many clicks.

[0182] when >10 o'clock, The closer it is to 1, the greater the increase in the interest score. In particular, when the push audience has almost completely browsed the page, the interest score will increase significantly. This relationship shows that the push audience who has completely browsed the page usually has a stronger interest in the content. This ensures that the effect accelerates. If the audience has viewed most or all of the content, their interest in the page is likely high, so that portion increases the interest score. The power function makes this effect more prominent as the page is closer to completion.

[0183] Use mathematical analysis methods to analyze historical viewing completeness, historical click-through rate, historical bounce rate, and historical browsing time to obtain historical interest levels. Subtract the historical interest level from the current interest level to obtain a degree subtraction value. Obtain the interest level error range from the push management system, match the degree subtraction value with the interest level error range, and determine whether the degree subtraction value is within the interest level error range. If not, it means that there is a large difference in the attention behavior of the push audience to the current content and historical content. Subtract the historical browsing time, historical bounce rate, historical click rate, and historical viewing completeness from the browsing time, bounce rate, click rate, and viewing completeness, respectively, to obtain the browsing time difference, bounce rate difference, click rate difference, and viewing completeness difference. The browsing time difference, bounce rate difference, click rate difference, and viewing completeness difference are determined as behavioral difference information.

[0184] Through this solution, the depth and breadth of the push audience's interaction with the current content can be quantified based on key indicators such as bounce rate, click-through rate and viewing completeness. By comparing the current content with the historical content type, if they are consistent, a comprehensive formula is used to combine factors such as browsing time, bounce rate, click-through rate and viewing completeness to accurately calculate the current interest of the push audience in the current content. At the same time, refer to historical attention information to determine whether the push audience's interests have changed significantly. If it exceeds the preset error range, further analyze the specific differences between historical and current browsing behaviors, including the difference in browsing time, bounce rate, click-through rate and viewing completeness. This is used as behavioral difference information to provide a scientific basis for the adjustment and optimization of precise push strategies, ensuring that the push content is more in line with the actual interests and needs of the push audience.

[0185] In some embodiments, the comprehensive difference index is determined based on the browsing time difference, the bounce rate difference, the click rate difference and the viewing completeness difference according to formula (2):

[0186] (2)

[0187] in, represents the comprehensive difference index, Indicates the preset browsing time coefficient. Indicates the browsing time difference. Indicates the preset bounce rate coefficient. Indicates the bounce rate difference. Indicates the preset click rate coefficient, Indicates the difference in click rate, Indicates the preset completeness coefficient, Indicates the viewing completeness difference;

[0188] According to the comprehensive difference index, the current attention time and historical attention time of the push audience to the current content are determined; the current attention time, historical attention time and active time period are determined as dynamic behavior information.

[0189] The comprehensive difference index may be a numerical value that measures the difference between the attention behaviors of the push audience towards the current content and the historical content.

[0190] The current attention time may be the length of time that the push audience pays attention to the current content.

[0191] The historical attention time can be the length of time that the push audience pays attention to the historical content.

[0192] The preset browsing time coefficient may be a value corresponding to the influence of the browsing time difference on the calculation result when calculating the comprehensive difference index.

[0193] The preset bounce rate coefficient may be a value corresponding to a measure of the impact of the bounce rate difference on the calculation result when calculating the comprehensive difference index.

[0194] The preset click rate coefficient may be a value corresponding to the impact of the click rate difference on the calculation result when calculating the comprehensive difference index.

[0195] The preset completeness coefficient may be a value corresponding to a measure of the impact of the viewing completeness difference on the calculation result when calculating the comprehensive difference index.

[0196] The active period may be a relatively fixed time period when the push audience pays attention to the content.

[0197] Specifically, before the push audience browses the push content, that is, before the push content is pushed to the push audience, there will be a large difference between the push audience's several operational behavior characteristics for historical content and current content, which makes it impossible to accurately determine the push audience's specific operational behavior information for historical content and current content. Therefore, it is necessary to calculate the difference between the push audience's several operational behavior characteristics for historical content and current content, so as to determine the comprehensive difference characteristics of the push audience's operational behavior for historical content and current content, and provide an accurate reference basis for the determination of subsequent dynamic behavior information.

[0198] Obtain historical difference information from the push management system, extract historical browsing time difference, historical bounce rate difference, historical comprehensive difference index, historical click rate difference and historical viewing completeness difference from the historical difference information, use linear regression analysis algorithm to analyze historical browsing time difference, historical bounce rate difference, historical comprehensive difference index, historical click rate difference and historical viewing completeness difference, and obtain preset browsing time coefficient, preset bounce rate coefficient, preset click rate coefficient and preset completeness coefficient. Use mathematical analysis algorithm to analyze browsing time difference, bounce rate difference, click rate difference, viewing completeness difference, preset browsing time coefficient, preset bounce rate coefficient, preset click rate coefficient and preset completeness coefficient, and calculate according to formula (2) to obtain comprehensive difference index. In formula (2), when When it is larger, it means that the browsing time of the current content is significantly better than that of the historical content, which has an impact on the comprehensive difference index. The contribution of Determines the impact of browsing time difference on the final index If Smaller (larger negative value), indicating that the bounce rate is significantly reduced, the current content is more attractive to the push audience, and the comprehensive difference index The positive contribution of Controlling the bounce rate difference The greater the difference in click rate, the more attractive the content is. The greater the positive contribution of Controls the impact of click-through rate difference in the final index. View completion difference The larger the value, the more attractive and deeply involved the content is to the audience, which has a greater impact on the comprehensive difference index. The greater the positive contribution of Control the impact of the completeness difference in the final index. According to the comprehensive difference index, determine the current attention time and historical attention time of the push audience to the current content, and determine the current attention time, historical attention time and active period as dynamic behavior information.

[0199] Through this solution, by combining the browsing time difference, bounce rate difference, click rate difference and viewing completeness difference, and assigning each difference with a preset coefficient, a comprehensive difference index is calculated, which effectively measures the comprehensive difference between the current and historical behaviors of the push audience. Based on the analysis of the comprehensive difference index, the current attention time and historical attention time of the push audience on the current content are accurately determined. The comparison of these two time points reveals the changing trend of the push audience's interest in browsing content. Combining the current attention time, historical attention time and the active period of the push audience forms complete dynamic behavior information, which provides strong data support for in-depth understanding of the push audience's behavior patterns and optimization of push strategies, and helps to improve the accuracy of content push and the push audience's participation.

[0200] In some embodiments, the push frequency is determined based on the push type, push audience, current interest level and search type; the push time period is determined based on the active time period and current attention time; the push balance information is determined based on the historical attention time, current attention time, historical content and current content; the push frequency, push time period and push balance information are determined as push information.

[0201] The search type may be the type of relevant content that the push audience corresponds to when searching for content.

[0202] The push period may be a time period determined when the push content is pushed to the push recipient.

[0203] The push frequency may be the number of times the push content is pushed to the push recipient in a unit of time.

[0204] The push balance information may be information corresponding to the duration determined for pushing the push content to the push recipients within a day and the duration of pushing the historical content.

[0205] Specifically, when the push audience pays attention to the content, since the push audience's active time period, content search type and attention duration are changing in real time, it is necessary to determine the push frequency, time period and other information of the push content to the push audience based on the push audience's real-time changing operational behavior on the content, so as to make the push process of the push content more in line with the actual push needs of the push audience.

[0206] Based on the push audience, determine whether the push type is consistent with the search type. If the push type is consistent with the search type, then before determining the push frequency, pre-set an interest level and its corresponding push frequency to obtain a preset level frequency set. Match the current interest level with the preset level frequency set to obtain a push frequency that matches the current interest level. Based on the current attention time, determine the active period as the push period. Based on the historical content and the current content, determine the historical attention time and the current attention time as the push duration of the historical content and the push duration of the current content, respectively, and determine the push duration of the historical content and the push duration of the current content as the push balance information. Determine the push frequency, push period and push balance information as push information.

[0207] Through this solution, the push frequency is accurately determined through intelligent algorithms to ensure that the pushed content neither disturbs the push audience too frequently nor misses the best display opportunity due to sparseness. At the same time, according to the active time period and current attention time of the push audience, the push time period is accurately set so that the push content can be presented at the time point when the push audience is most likely to pay attention, thereby improving the effectiveness of the push content. In addition, combined with a comprehensive analysis of historical attention time, current attention time, historical content and current content, push balance information is formulated to maintain the freshness and coherence of the push content and avoid repetition or disconnection among the push audience. The push frequency, push time period and push balance information are integrated into push information to achieve comprehensive optimization of push information, ensure that the push of push content is both efficient and personalized, and effectively improve the satisfaction and participation of the push audience.

[0208] In some embodiments, based on the current content and the historical content, the current content type corresponding to the current content and the historical content type corresponding to the historical content are determined; based on the current content type and the historical content type, the degree of content similarity is determined; based on the degree of content similarity, it is determined whether the push type and the search type are consistent; if the push type and the search type are consistent, the interest level is obtained, the current interest level is matched with the interest level, and a push frequency corresponding to the current interest level is obtained.

[0209] The content similarity may be a numerical value indicating the consistency between the current content and the historical content.

[0210] The interest level level may be a numerical level that measures the degree of preference of the push audience for the content they are interested in.

[0211] Specifically, when the push audience pays attention to content, there will be certain similarities between the push audience's search types for historical content and the search types for current content. In order to make the push of push content more in line with the frequency needs of the push audience in different time periods, it is necessary to consider the push audience's attention information on historical content and current content as well as the push audience's own information, so as to better meet the push audience's frequency needs for push content in different time periods.

[0212] Based on the type of the current content and the type of the historical content obtained in the above steps, the current content and the historical content are analyzed using text vectorization technology to obtain the content similarity between the current content and the historical content. Based on the content similarity obtained in the above steps, it is determined whether the push type is consistent with the search type. If the push type is consistent with the search type, the current interest level is matched with the interest level to obtain the push frequency corresponding to the current interest level, and the push frequency corresponding to the current interest level is determined as the push frequency.

[0213] Through this solution, by comparing and analyzing the current content and historical content, the types of each are clarified, and the similarity of the content between the two is evaluated accordingly, providing a basis for the formulation of subsequent push strategies. Based on the similarity of the content, it is determined whether the push type is consistent with the search type to ensure that the push content is highly consistent with the push audience's intention. If consistent, the interest level system is further introduced to match the current interest level with the interest level level, and the push frequency is determined based on this to ensure that the push intensity matches the push audience's interest in the push content. While ensuring the push audience experience, it effectively improves the visibility and conversion rate of push information, and realizes the seamless connection between the precise push of push content and the push audience's demand for push content.

[0214] In some embodiments, according to the active period and the current attention time, a number of time distribution nodes at which the push audience pays attention to the current content during the active period are determined; a number of time distribution coordinates of the number of time distribution nodes in a day and the total browsing time of the push audience in a day are obtained, and according to the total browsing time and the current attention time, the attention intensity of the push audience to the current content is determined; according to the number of time distribution coordinates, attention intensity, total browsing time of a day and the active period, the node time ratio corresponding to each time distribution node is calculated according to formula (3):

[0215] (3)

[0216] in, Indicates The active duration of each time node, Indicates the intensity of attention, Indicates The location of the time node, Indicates the total length of a day;

[0217] Get the focus position of the push audience when they pay attention to the current content, and determine the push time period based on the node time ratio and the focus position.

[0218] The time distribution node can be the position where the time the push audience pays attention to the current content is distributed within the total time of a day. The time distribution node can be morning, noon, afternoon or evening.

[0219] The time distribution coordinates may be corresponding moments of the time distribution nodes within the total duration of a day.

[0220] The attention strength may be a numerical value indicating the degree of attention paid by the push audience to the current content.

[0221] The node time ratio may be a ratio of the time that the push audience pays attention to the current content within one day to the total time of the day.

[0222] The attention location may be the physical location where the push audience is located when paying attention to the current content.

[0223] Specifically, when the push audience pays attention to the content, the actual environment corresponding to the push audience's physical location and the length of time they pay attention to the current content will affect the push audience's attention intensity to a certain extent, thereby affecting the determination of the push period. Therefore, it is necessary to analyze the push audience's physical location and the corresponding attention duration to determine the push period that meets the actual push needs of the push audience.

[0224] Based on the current attention time determined in the above steps, the active time period is matched with the total time of the day to obtain several positions of the active time period within the total time of the day, and the several positions of the active time period within the total time of the day are determined as several time distribution nodes. The several time distribution nodes obtained in the above steps are analyzed using time series analysis technology to obtain several time distribution coordinates of the several time distribution nodes within the day. The browsing time of all content by the push audience in one day is obtained from the push management system, and the browsing time of all content by the push audience in one day is determined as the total browsing time. The current attention time is divided by the total browsing time to obtain the attention intensity of the push audience to the current content. The several time distribution coordinates, attention intensity, total time of the day and active time period are analyzed using a mathematical analysis method, and the node time ratio corresponding to each time distribution node is calculated according to formula (3). The position of the push audience's device when paying attention to the current content is obtained from the push management system, and the position of the push audience's device when paying attention to the current content is determined as the attention position. The humanistic information corresponding to the attention position is obtained from the humanistic information website, and the push content corresponding to the humanistic information corresponding to the attention position is obtained from the push management system. Obtain the duration corresponding to the focus location from the push management system, subtract the duration corresponding to the focus location from the total duration of the day to obtain the duration corresponding to the pushed content. Multiply the node time ratio by the duration corresponding to the pushed content to obtain the total push duration of the pushed content in one day. Based on the total push duration, match several time distribution nodes with the total duration of the day to obtain several time periods, and determine the several time periods as push periods.

[0225] Through this solution, by identifying several time distribution nodes, the audience's attention timeline can be accurately depicted. Furthermore, by combining the time distribution coordinates of these nodes in a day and the total browsing time of the push audience, the push audience's attention intensity to the current content is calculated, which effectively measures the push audience's interest and involvement in the current content. Using information such as attention intensity, time node location, total duration of a day, and active time period, the active time ratio of each time node is calculated through a specific formula, reflecting the activity level of the push audience at different time points. Combined with the location information of the push audience when paying attention and the node time ratio, the push time period is accurately determined to ensure that the push information can be presented at the time point when the push audience is most active and concerned, thereby significantly improving the push effect and push audience participation.

[0226] In some embodiments, the first consumed time corresponding to the historical content is determined based on the historical attention time and the historical content; the current consumed time corresponding to the current content is determined based on the current attention time and the current content; the total current push time of the current content within one day and the total historical push time of the historical content within one day are determined based on the consumed time, the first consumed time and the current consumed time, and the current push time and the historical push time are determined as push balance information.

[0227] The first elapsed time may be the length of time it takes for the push audience to start searching for historical content and end paying attention to it.

[0228] The current consumed time may be the length of time that the push audience consumes from the beginning of searching to the end of paying attention to the current content.

[0229] The current push duration may be a time length determined when the current content is pushed to the push audience within one day.

[0230] The historical push duration may be the length of time determined when pushing historical content to a push audience within a day.

[0231] Specifically, when the push audience pays attention to content, the content that the push audience pays attention to is different at different times of the day. In order to determine the push duration of historical content and current content in a day respectively, it is necessary to consider the attention information of the push audience to different content at different times of the day to determine the corresponding push duration of different content within a day.

[0232] Based on the historical content obtained in the above steps, the historical attention time is determined as the first consumption time. Based on the current content obtained in the above steps, the current attention time is determined as the current consumption time. The current consumption time and the first consumption time are respectively determined as the current push duration and the historical push duration. The current push duration and the historical push duration are determined as push balance information.

[0233] Through this solution, based on the historical attention time and historical content, the first consumption time consumed by the historical content is calculated, reflecting the degree of attention of the push audience to similar content in the past. Combined with the current attention time and the current content, the current consumption time corresponding to the current content is determined, which effectively measures the push audience's immediate interest and attention to the current content. By comparing the historical consumption time, the first consumption time and the current consumption time, the total push time of the current content in a day and the total push time of the historical content in a day are comprehensively evaluated to ensure the balance of the push content and avoid the push audience from feeling fatigued or ignoring other important information due to excessive reception of similar information. Using the current push time and the historical push time as push balance information provides an important basis for formulating scientific and reasonable push strategies, which helps to improve the satisfaction of the push audience and the dissemination effect of the push content.

[0234] In some embodiments, the number of views and first viewing completeness of several contents corresponding to the current content type by the push audience after the push is obtained; based on the number of views and first viewing completeness corresponding to each content, the same feature information of the several contents is determined; a number of dwell times and a number of click-through rates of the several contents corresponding to the same feature information are obtained; based on the number of dwell times and the number of click-through rates, the degree of attraction of the several contents corresponding to the same feature information to the push audience is determined; based on the degree of attraction, it is determined whether the push audience has a long-term interest in the several contents corresponding to the same feature information.

[0235] The first viewing completeness may be a ratio of the content portion that the push audience has followed to the total content portion when following the content corresponding to the current content type.

[0236] The degree of attraction may be the degree to which the push audience pays attention to the content corresponding to the current content type.

[0237] The long-term interest may be the long-term attention interest that the push audience has on the content corresponding to the current content type.

[0238] The same characteristic information may be information corresponding to the same characteristic of several contents corresponding to the current content type.

[0239] Specifically, when the push audience is interested in the current content, there is a possibility that the push audience will develop a long-term interest in the current content. At this time, it is necessary to analyze and judge the push audience's operational behavior information on the current content to determine whether the push audience's interest in the current content will evolve into a long-term behavioral interest.

[0240] The number of views and the first viewing completeness of the push audience for the current content type corresponding to the current content type are obtained from the push management system. Based on the number of views and the first viewing completeness obtained in the above steps, the feature recognition algorithm is used to analyze the several contents to obtain the same feature information of the several contents. The several stay times and several click rates of the push audience on the several contents are obtained from the push management system. The several stay times and several click rates obtained in the above steps are analyzed using a mathematical analysis algorithm to obtain the attraction degree. A level of attraction is pre-set to obtain a preset level of attraction, and the preset level of attraction is divided into high, medium and low. The attraction degree obtained in the above steps is matched with the preset level of attraction. If the attraction degree is high or medium, it is determined that the push audience has a long-term interest in the several contents corresponding to the same feature information. If the attraction degree is low, it means that the attraction degree of the several contents corresponding to the same feature information to the push audience does not meet the standard for determining long-term interest. At this time, it is determined that the push audience has no long-term interest in the several contents corresponding to the same feature information.

[0241] Through this solution, by comprehensively tracking the browsing behavior of push audiences on content related to the current content type after receiving the push content, including the number of views and the completeness of the first viewing, the push audience's acceptance of these contents is deeply analyzed. Furthermore, based on the same feature information extracted from the browsing behavior, the dwell time and click-through rate of the corresponding content are recorded in detail. These two indicators together reflect the attractiveness of the content to the push audience. Through the comprehensive analysis of these data, it is accurately assessed which features of content have continued appeal to the push audience, so as to determine whether the push audience has a long-term interest in these feature contents. It not only helps to optimize the selection and strategy of push content, but also provides valuable feedback to content creators, promotes continuous innovation and improvement of push content, and meets the long-term interests and needs of push audiences.

[0242] Figure 3 A schematic diagram of a traffic intelligent management system provided in one embodiment of the present application is shown in FIG. Figure 3 As shown, the intelligent traffic management system 300 of this embodiment includes: an audience information determination module 301, a behavior information determination module 302 and a push information determination module 303.

[0243] The audience information determination module 301 is used to obtain push content and determine push audience information according to the push content;

[0244] The behavior information determination module 302 is used to obtain the behavior information of the push audience, and determine the dynamic behavior information according to the push audience information and the behavior information;

[0245] The push information determination module 303 is used to obtain the current content information of the push audience, and determine the push information according to the push audience information, the dynamic behavior information and the current content information.

[0246] Optionally, the behavior information determination module 302 is specifically configured to:

[0247] Determining a push type of the pushed content according to the pushed content;

[0248] Acquire the real-time behavior information and historical behavior information of the push recipient, and determine behavior difference information based on the real-time behavior information and the historical behavior information;

[0249] The dynamic behavior information of the push recipient is determined according to the behavior difference information, the push type and the push recipient information.

[0250] Optionally, the behavior information determination module 302 is specifically configured to:

[0251] Determining the bounce rate of the push audience for the current content according to the bounce information;

[0252] Determining the click rate of the push audience for the current content according to the click information;

[0253] Determining the viewing completeness of the current content by the push audience according to the viewing information;

[0254] Obtaining historical content and current content, and determining whether the current content and the historical content are of the same type;

[0255] If the current content and the historical content are of the same type, the current interest level of the push audience for the current content is determined based on the browsing time, the bounce rate, the click rate, and the viewing completeness, and is calculated according to the following formula:

[0256] ;

[0257] in, represents the current level of interest, Indicates the preset time weight, Indicates the browsing time. Indicates the preset bounce rate weight. represents the bounce rate, Indicates the preset click rate weight. represents the click-through rate, Indicates the preset completeness weight, Indicates the viewing completeness. Indicates preset adjustment constant;

[0258] Determining the historical interest level of the push recipient in the historically-focused content according to the historically-focused information;

[0259] Obtaining an interest level error range, and determining whether a difference between the historical interest level and the current interest level is within the interest level error range;

[0260] If the difference between the historical interest level and the current interest level is not within the interest level error range, determining the historical browsing time, historical bounce rate, historical click rate and historical viewing completeness according to the historical attention information;

[0261] Determine a browsing time difference, a bounce rate difference, a click rate difference and a viewing completeness difference according to the historical browsing time, the historical bounce rate, the historical click rate, the historical viewing completeness, the browsing time, the bounce rate, the click rate and the viewing completeness;

[0262] The browsing time difference, the bounce rate difference, the click rate difference and the viewing completeness difference are determined as the behavior difference information.

[0263] Optionally, the behavior information determination module 302 is specifically configured to:

[0264] According to the browsing time difference, the bounce rate difference, the click rate difference and the viewing completeness difference, a comprehensive difference index is determined and calculated according to the following formula:

[0265] ;

[0266] in, represents the comprehensive difference index, Indicates the preset browsing time coefficient. represents the browsing time difference, Indicates the preset bounce rate coefficient. represents the bounce rate difference, Indicates the preset click rate coefficient, represents the click rate difference, Indicates the preset completeness coefficient, represents the viewing completeness difference;

[0267] Determining the current attention time and historical attention time of the push audience to the current content according to the comprehensive difference index;

[0268] The current attention time, the historical attention time and the active time period are determined as the dynamic behavior information.

[0269] Optionally, the push information determination module 303 is specifically used to:

[0270] Determine the push frequency according to the push type, the push recipient, the current interest level and the search type;

[0271] Determine a push time period according to the active time period and the current attention time;

[0272] Determine push balance information according to the historical attention time, the current attention time, the historical content and the current content;

[0273] The push frequency, the push time period, and the push balance information are determined as the push information.

[0274] Optionally, the push information determination module 303 is specifically used to:

[0275] Determining, according to the current content and the historical content, a current content type corresponding to the current content and a historical content type corresponding to the historical content;

[0276] Determining a content similarity degree according to the current content type and the historical content type;

[0277] According to the similarity of the contents, determining whether the push type is consistent with the search type;

[0278] If the push type and the search type are consistent, the interest level level is obtained, the current interest level is matched with the interest level level, and a push frequency corresponding to the current interest level is obtained.

[0279] Optionally, the push information determination module 303 is specifically used to:

[0280] According to the active time period and the current attention time, determining a number of time distribution nodes at which the push audience pays attention to the current content within the active time period;

[0281] Acquire a number of time distribution coordinates of the number of time distribution nodes in one day and the total browsing time of the push audience in one day, and determine the attention intensity of the push audience to the current content according to the total browsing time and the current attention time;

[0282] According to the time distribution coordinates, the attention intensity, the total duration of a day and the active period, the node time ratio corresponding to each time distribution node is determined and calculated according to the following formula:

[0283] ;

[0284] in, Indicates The active duration of each time node, represents the intensity of attention, Indicates The location of the time node, represents the total length of the day in question;

[0285] The focus position of the push audience when paying attention to the current content is obtained, and the push time period is determined according to the node time ratio and the focus position.

[0286] Optionally, the push information determination module 303 is specifically used to:

[0287] Determining a first consumed time corresponding to the historical content according to the historical attention time and the historical content;

[0288] Determine, according to the current attention time and the current content, the current consumption time corresponding to the current content;

[0289] According to the consumed time, the first consumed time and the current consumed time, determine the total current push time of the current content within one day and the total historical push time of the historical content within one day, and determine the current push time and the historical push time as the push balance information.

[0290] Optionally, the intelligent traffic management system 300 further includes a long-term interest determination module 304, which is specifically used to:

[0291] Obtaining the number of views and first viewing completeness of a number of contents corresponding to the current content type by the push audience after the push;

[0292] Determining the same characteristic information of the plurality of contents according to the number of views corresponding to each content and the first viewing completeness;

[0293] Obtaining a number of dwell times and a number of click rates of a number of contents corresponding to the same characteristic information;

[0294] Determining, based on the plurality of dwell times and the plurality of click rates, the attractiveness of the plurality of contents corresponding to the same characteristic information to the push audience;

[0295] According to the attraction degree, it is determined whether the push recipient has a long-term interest in the plurality of contents corresponding to the same characteristic information.

[0296] The system of this embodiment can be used to execute the method of any of the above embodiments. The implementation principles and technical effects are similar and will not be described in detail here.

Claims

1. A method for intelligent traffic management, characterized in that: include: Obtain push content, and determine push recipient information based on the push content; Acquire behavior information of a push recipient, and determine dynamic behavior information based on the push recipient information and the behavior information; Acquire current content information of the push recipient, and determine the push information according to the push recipient information, the dynamic behavior information and the current content information; The behavior information includes real-time behavior information and historical behavior information. The dynamic behavior information is determined based on the behavior information and the push audience information, including: Determining a push type of the pushed content according to the pushed content; Determining behavior difference information according to the real-time behavior information and the historical behavior information; Determine the dynamic behavior information of the push recipient according to the behavior difference information, the push type and the push recipient information; The real-time behavior information includes browsing time, pop-up information, click information and viewing information of the push audience for the current content, and the historical behavior information includes historical attention information. The behavior difference information is determined based on the real-time behavior information and the historical behavior information, including: Determining the bounce rate of the push audience for the current content according to the bounce information; Determining the click rate of the push audience for the current content according to the click information; Determining the viewing completeness of the current content by the push audience according to the viewing information; Obtaining historical content and current content, and determining whether the current content and the historical content are of the same type; If the current content and the historical content are of the same type, the current interest level of the push audience for the current content is determined based on the browsing time, the bounce rate, the click rate, and the viewing completeness, and is calculated according to the following formula: ； in, represents the current level of interest, Indicates the preset time weight, Indicates the browsing time. Indicates the preset bounce rate weight. represents the bounce rate, Indicates the preset click rate weight. represents the click-through rate, Indicates the preset completeness weight, Indicates the viewing completeness. Indicates preset adjustment constant; Determining the historical interest level of the push recipient in the historically-focused content according to the historically-focused information; Obtaining an interest level error range, and determining whether a difference between the historical interest level and the current interest level is within the interest level error range; If the difference between the historical interest level and the current interest level is not within the interest level error range, determining the historical browsing time, historical bounce rate, historical click rate and historical viewing completeness according to the historical attention information; Determine a browsing time difference, a bounce rate difference, a click rate difference and a viewing completeness difference according to the historical browsing time, the historical bounce rate, the historical click rate, the historical viewing completeness, the browsing time, the bounce rate, the click rate and the viewing completeness; Determine the browsing time difference, the bounce rate difference, the click rate difference and the viewing completeness difference as the behavior difference information; The push audience information includes the active time period of the push audience, and determining the dynamic behavior information of the push audience according to the behavior difference information, the push type and the push audience information includes: According to the browsing time difference, the bounce rate difference, the click rate difference and the viewing completeness difference, a comprehensive difference index is determined and calculated according to the following formula: ； in, represents the comprehensive difference index, Indicates the preset browsing time coefficient. represents the browsing time difference, Indicates the preset bounce rate coefficient. represents the bounce rate difference, Indicates the preset click rate coefficient, represents the click rate difference, Indicates the preset completeness coefficient, represents the viewing completeness difference; Determining the current attention time and historical attention time of the push audience to the current content according to the comprehensive difference index; The current attention time, the historical attention time and the active time period are determined as the dynamic behavior information.

2. The method according to claim 1, characterized in that The current content information includes the search type of the current content, and the determining the push information according to the push audience information, the dynamic behavior information and the current content information includes: Determine the push frequency according to the push type, the push recipient, the current interest level and the search type; Determine a push time period according to the active time period and the current attention time; Determine push balance information according to the historical attention time, the current attention time, the historical content and the current content; The push frequency, the push time period, and the push balance information are determined as the push information.

3. The method according to claim 2, characterized in that The determining of the push frequency according to the push type, the push recipient, the interest level and the search type includes: Determining, according to the current content and the historical content, a current content type corresponding to the current content and a historical content type corresponding to the historical content; Determining a content similarity degree according to the current content type and the historical content type; According to the similarity of the contents, determining whether the push type is consistent with the search type; If the push type and the search type are consistent, the interest level level is obtained, the current interest level is matched with the interest level level, and a push frequency corresponding to the current interest level is obtained.

4. The method according to claim 2, characterized in that: The determining of the push time period according to the active time period and the current attention time includes: According to the active time period and the current attention time, determining a number of time distribution nodes at which the push audience pays attention to the current content within the active time period; Acquire a number of time distribution coordinates of the number of time distribution nodes in one day and the total browsing time of the push audience in one day, and determine the attention intensity of the push audience to the current content according to the total browsing time and the current attention time; According to the time distribution coordinates, the attention intensity, the total duration of a day and the active period, the node time ratio corresponding to each time distribution node is determined and calculated according to the following formula: ； in, Indicates The active duration of each time node, represents the intensity of attention, Indicates The location of the time node, represents the total length of the day in question; The focus position of the push audience when paying attention to the current content is obtained, and the push time period is determined according to the node time ratio and the focus position.

5. The method according to claim 2, characterized in that: The determining push balance information according to the historical attention time, the current attention time, the historical content and the current content includes: Determining a first consumed time corresponding to the historical content according to the historical attention time and the historical content; Determine, according to the current attention time and the current content, the current consumption time corresponding to the current content; According to the consumed time, the first consumed time and the current consumed time, determine the total current push time of the current content within one day and the total historical push time of the historical content within one day, and determine the current push time and the historical push time as the push balance information.

6. The method according to claim 5, characterized in that After determining the dynamic behavior information of the push audience according to the behavior difference information and the push audience information, the method further includes: Obtaining the number of views and first viewing completeness of a number of contents corresponding to the current content type by the push audience after the push; Determining the same characteristic information of the plurality of contents according to the number of views corresponding to each content and the first viewing completeness; Obtaining a number of dwell times and a number of click rates of a number of contents corresponding to the same characteristic information; Determining, based on the plurality of dwell times and the plurality of click rates, the attractiveness of the plurality of contents corresponding to the same characteristic information to the push audience; According to the attraction degree, it is determined whether the push recipient has a long-term interest in the plurality of contents corresponding to the same characteristic information.

7. A traffic intelligent management system, characterized in that: The method as claimed in any one of claims 1 to 6 comprises: An audience information determination module is used to obtain push content and determine push audience information based on the push content; A behavior information determination module, used to obtain the behavior information of the push recipient, and determine the dynamic behavior information according to the push recipient information and the behavior information; The push information determination module is used to obtain the current content information of the push audience, and determine the push information according to the push audience information, the dynamic behavior information and the current content information.

Images